summaryrefslogtreecommitdiff
path: root/kernel/time
diff options
context:
space:
mode:
Diffstat (limited to 'kernel/time')
-rw-r--r--kernel/time/clocksource.c11
-rw-r--r--kernel/time/hrtimer.c11
-rw-r--r--kernel/time/timekeeping.c6
-rw-r--r--kernel/time/timekeeping_internal.h8
-rw-r--r--kernel/time/timer_migration.c64
5 files changed, 83 insertions, 17 deletions
diff --git a/kernel/time/clocksource.c b/kernel/time/clocksource.c
index aab6472853fa..7304d7cf47f2 100644
--- a/kernel/time/clocksource.c
+++ b/kernel/time/clocksource.c
@@ -24,7 +24,7 @@ static void clocksource_enqueue(struct clocksource *cs);
static noinline u64 cycles_to_nsec_safe(struct clocksource *cs, u64 start, u64 end)
{
- u64 delta = clocksource_delta(end, start, cs->mask);
+ u64 delta = clocksource_delta(end, start, cs->mask, cs->max_raw_delta);
if (likely(delta < cs->max_cycles))
return clocksource_cyc2ns(delta, cs->mult, cs->shift);
@@ -993,6 +993,15 @@ static inline void clocksource_update_max_deferment(struct clocksource *cs)
cs->max_idle_ns = clocks_calc_max_nsecs(cs->mult, cs->shift,
cs->maxadj, cs->mask,
&cs->max_cycles);
+
+ /*
+ * Threshold for detecting negative motion in clocksource_delta().
+ *
+ * Allow for 0.875 of the counter width so that overly long idle
+ * sleeps, which go slightly over mask/2, do not trigger the
+ * negative motion detection.
+ */
+ cs->max_raw_delta = (cs->mask >> 1) + (cs->mask >> 2) + (cs->mask >> 3);
}
static struct clocksource *clocksource_find_best(bool oneshot, bool skipcur)
diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
index 80fe3749d2db..030426c8c944 100644
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -2202,6 +2202,15 @@ int hrtimers_prepare_cpu(unsigned int cpu)
}
cpu_base->cpu = cpu;
+ hrtimer_cpu_base_init_expiry_lock(cpu_base);
+ return 0;
+}
+
+int hrtimers_cpu_starting(unsigned int cpu)
+{
+ struct hrtimer_cpu_base *cpu_base = this_cpu_ptr(&hrtimer_bases);
+
+ /* Clear out any left over state from a CPU down operation */
cpu_base->active_bases = 0;
cpu_base->hres_active = 0;
cpu_base->hang_detected = 0;
@@ -2210,7 +2219,6 @@ int hrtimers_prepare_cpu(unsigned int cpu)
cpu_base->expires_next = KTIME_MAX;
cpu_base->softirq_expires_next = KTIME_MAX;
cpu_base->online = 1;
- hrtimer_cpu_base_init_expiry_lock(cpu_base);
return 0;
}
@@ -2286,5 +2294,6 @@ int hrtimers_cpu_dying(unsigned int dying_cpu)
void __init hrtimers_init(void)
{
hrtimers_prepare_cpu(smp_processor_id());
+ hrtimers_cpu_starting(smp_processor_id());
open_softirq(HRTIMER_SOFTIRQ, hrtimer_run_softirq);
}
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 0ca85ff4fbb4..3d128825d343 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -755,7 +755,8 @@ static void timekeeping_forward_now(struct timekeeper *tk)
u64 cycle_now, delta;
cycle_now = tk_clock_read(&tk->tkr_mono);
- delta = clocksource_delta(cycle_now, tk->tkr_mono.cycle_last, tk->tkr_mono.mask);
+ delta = clocksource_delta(cycle_now, tk->tkr_mono.cycle_last, tk->tkr_mono.mask,
+ tk->tkr_mono.clock->max_raw_delta);
tk->tkr_mono.cycle_last = cycle_now;
tk->tkr_raw.cycle_last = cycle_now;
@@ -2230,7 +2231,8 @@ static bool timekeeping_advance(enum timekeeping_adv_mode mode)
return false;
offset = clocksource_delta(tk_clock_read(&tk->tkr_mono),
- tk->tkr_mono.cycle_last, tk->tkr_mono.mask);
+ tk->tkr_mono.cycle_last, tk->tkr_mono.mask,
+ tk->tkr_mono.clock->max_raw_delta);
/* Check if there's really nothing to do */
if (offset < real_tk->cycle_interval && mode == TK_ADV_TICK)
diff --git a/kernel/time/timekeeping_internal.h b/kernel/time/timekeeping_internal.h
index 63e600e943a7..8c9079108ffb 100644
--- a/kernel/time/timekeeping_internal.h
+++ b/kernel/time/timekeeping_internal.h
@@ -30,15 +30,15 @@ static inline void timekeeping_inc_mg_floor_swaps(void)
#endif
-static inline u64 clocksource_delta(u64 now, u64 last, u64 mask)
+static inline u64 clocksource_delta(u64 now, u64 last, u64 mask, u64 max_delta)
{
u64 ret = (now - last) & mask;
/*
- * Prevent time going backwards by checking the MSB of mask in
- * the result. If set, return 0.
+ * Prevent time going backwards by checking the result against
+ * @max_delta. If greater, return 0.
*/
- return ret & ~(mask >> 1) ? 0 : ret;
+ return ret > max_delta ? 0 : ret;
}
/* Semi public for serialization of non timekeeper VDSO updates. */
diff --git a/kernel/time/timer_migration.c b/kernel/time/timer_migration.c
index 8d57f7686bb0..066c9ddca4ec 100644
--- a/kernel/time/timer_migration.c
+++ b/kernel/time/timer_migration.c
@@ -534,8 +534,13 @@ static void __walk_groups(up_f up, struct tmigr_walk *data,
break;
child = group;
- group = group->parent;
+ /*
+ * Pairs with the store release on group connection
+ * to make sure group initialization is visible.
+ */
+ group = READ_ONCE(group->parent);
data->childmask = child->groupmask;
+ WARN_ON_ONCE(!data->childmask);
} while (group);
}
@@ -564,7 +569,7 @@ static struct tmigr_event *tmigr_next_groupevt(struct tmigr_group *group)
while ((node = timerqueue_getnext(&group->events))) {
evt = container_of(node, struct tmigr_event, nextevt);
- if (!evt->ignore) {
+ if (!READ_ONCE(evt->ignore)) {
WRITE_ONCE(group->next_expiry, evt->nextevt.expires);
return evt;
}
@@ -660,7 +665,7 @@ static bool tmigr_active_up(struct tmigr_group *group,
* lock is held while updating the ignore flag in idle path. So this
* state change will not be lost.
*/
- group->groupevt.ignore = true;
+ WRITE_ONCE(group->groupevt.ignore, true);
return walk_done;
}
@@ -721,6 +726,7 @@ bool tmigr_update_events(struct tmigr_group *group, struct tmigr_group *child,
union tmigr_state childstate, groupstate;
bool remote = data->remote;
bool walk_done = false;
+ bool ignore;
u64 nextexp;
if (child) {
@@ -739,11 +745,19 @@ bool tmigr_update_events(struct tmigr_group *group, struct tmigr_group *child,
nextexp = child->next_expiry;
evt = &child->groupevt;
- evt->ignore = (nextexp == KTIME_MAX) ? true : false;
+ /*
+ * This can race with concurrent idle exit (activate).
+ * If the current writer wins, a useless remote expiration may
+ * be scheduled. If the activate wins, the event is properly
+ * ignored.
+ */
+ ignore = (nextexp == KTIME_MAX) ? true : false;
+ WRITE_ONCE(evt->ignore, ignore);
} else {
nextexp = data->nextexp;
first_childevt = evt = data->evt;
+ ignore = evt->ignore;
/*
* Walking the hierarchy is required in any case when a
@@ -769,7 +783,7 @@ bool tmigr_update_events(struct tmigr_group *group, struct tmigr_group *child,
* first event information of the group is updated properly and
* also handled properly, so skip this fast return path.
*/
- if (evt->ignore && !remote && group->parent)
+ if (ignore && !remote && group->parent)
return true;
raw_spin_lock(&group->lock);
@@ -783,7 +797,7 @@ bool tmigr_update_events(struct tmigr_group *group, struct tmigr_group *child,
* queue when the expiry time changed only or when it could be ignored.
*/
if (timerqueue_node_queued(&evt->nextevt)) {
- if ((evt->nextevt.expires == nextexp) && !evt->ignore) {
+ if ((evt->nextevt.expires == nextexp) && !ignore) {
/* Make sure not to miss a new CPU event with the same expiry */
evt->cpu = first_childevt->cpu;
goto check_toplvl;
@@ -793,7 +807,7 @@ bool tmigr_update_events(struct tmigr_group *group, struct tmigr_group *child,
WRITE_ONCE(group->next_expiry, KTIME_MAX);
}
- if (evt->ignore) {
+ if (ignore) {
/*
* When the next child event could be ignored (nextexp is
* KTIME_MAX) and there was no remote timer handling before or
@@ -1487,6 +1501,21 @@ static void tmigr_init_group(struct tmigr_group *group, unsigned int lvl,
s.seq = 0;
atomic_set(&group->migr_state, s.state);
+ /*
+ * If this is a new top-level, prepare its groupmask in advance.
+ * This avoids accidents where yet another new top-level is
+ * created in the future and made visible before the current groupmask.
+ */
+ if (list_empty(&tmigr_level_list[lvl])) {
+ group->groupmask = BIT(0);
+ /*
+ * The previous top level has prepared its groupmask already,
+ * simply account it as the first child.
+ */
+ if (lvl > 0)
+ group->num_children = 1;
+ }
+
timerqueue_init_head(&group->events);
timerqueue_init(&group->groupevt.nextevt);
group->groupevt.nextevt.expires = KTIME_MAX;
@@ -1550,8 +1579,25 @@ static void tmigr_connect_child_parent(struct tmigr_group *child,
raw_spin_lock_irq(&child->lock);
raw_spin_lock_nested(&parent->lock, SINGLE_DEPTH_NESTING);
- child->parent = parent;
- child->groupmask = BIT(parent->num_children++);
+ if (activate) {
+ /*
+ * @child is the old top and @parent the new one. In this
+ * case groupmask is pre-initialized and @child already
+ * accounted, along with its new sibling corresponding to the
+ * CPU going up.
+ */
+ WARN_ON_ONCE(child->groupmask != BIT(0) || parent->num_children != 2);
+ } else {
+ /* Adding @child for the CPU going up to @parent. */
+ child->groupmask = BIT(parent->num_children++);
+ }
+
+ /*
+ * Make sure parent initialization is visible before publishing it to a
+ * racing CPU entering/exiting idle. This RELEASE barrier enforces an
+ * address dependency that pairs with the READ_ONCE() in __walk_groups().
+ */
+ smp_store_release(&child->parent, parent);
raw_spin_unlock(&parent->lock);
raw_spin_unlock_irq(&child->lock);
generated by cgit v1.2.3 (git 2.51.0) at 2025-09-18 03:39:14 +0000